AI data center project secretly sucked 29 million gallons of water over 15 months before detected by residents complaining about low water pressure

A data center project in Fayette County, Georgia, secretly consumed approximately 29 million gallons of water over 15 months before being detected by residents complaining about low water pressure. Th... Read original →

The 'Project Excalibur' water diversion scandal reveals a critical disconnect between AI data center cooling claims and municipal oversight. Despite using closed-loop systems, the 29-million-gallon unreported draw exposes systemic flaws in evaporative water use (EUW) modeling, accelerating industry shifts toward adiabatic or hybrid dry cooling—directly impacting liquid-cooled chip packaging and thermal interface material demand. Regulatory leniency due to tax reliance is unsustainable; Georgia’s worsening water stress will likely mandate real-time cloud metering with AI anomaly detection for new builds, raising CAPEX by 10–15%. Amazon and Google may leverage this to bolster their 'Water Positive' branding for ESG financing edge, while Microsoft faces reputational fallout via its QTS ties. Within 18 months, U.S. Southeast data center siting will pivot from 'low electricity cost' to 'water rights transparency,' triggering regional reallocation and intensifying direct competition between fabs and data centers for scarce water allocations.